Loading ...
Sorry, an error occurred while loading the content.
 

New network graphic / built with real world data.

Expand Messages
  • Mark
    Friends, Denis sent me some real world poll data to analyze. Of the 158 pollees, 32 participated in SD2 polling. Of those 32, 24 endorsed other participants. I
    Message 1 of 3 , Jun 1, 2005
      Friends,

      Denis sent me some real world poll data to analyze. Of the 158 pollees,
      32 participated in SD2 polling. Of those 32, 24 endorsed other
      participants.

      I put this data onto Excel and onto Ucinet. I had NetDraw/Ucinet draw
      the new network graphic and had the Newman-Girvan algorithm color the
      three main groups - you will see that the nodes are colored black, blue
      and red.

      This network is a bit sparse - it is not yet SD2 complient - and the
      centrality algorithms of Ucinet have not returned satisfactory results.
      This is not Denis's fault, and I predict that he will get more of the
      needed pollee participation in the future.

      Visone and PageRank have not yet been tried.

      Shanti
      Mark Rosst
      Seattle WA USA
    • denis bider
      Hi Mark, thank you for this image - it is interesting! ... Yet it seems somewhat clear to the human eye that nodes 7, 11 and 24 would likely have the highest
      Message 2 of 3 , Jun 1, 2005
        Hi Mark,

        thank you for this image - it is interesting!

        > This network is a bit sparse - it is not yet SD2 complient - and the
        > centrality algorithms of Ucinet have not returned satisfactory results.

        Yet it seems somewhat clear to the human eye that nodes 7, 11 and 24 would
        likely have the highest ranking (they receive the most input connections).
        Without the help of an algorithm, one would likely opt for these nodes as
        the leaders. Would that be a mistake?

        Can we look forward for more results based on analysis this data?
        Specifically, PageRank?

        Best regards!

        denis
      • Mark
        ... -M: Yes, thank you, it is interesting. ... the centrality algorithms of Ucinet have not returned satisfactory results. ... 24 would likely have the highest
        Message 3 of 3 , Jun 1, 2005
          --"denis bider" wrote:

          >D: Hi Mark, thank you for this image - it is interesting!

          -M: Yes, thank you, it is interesting.

          > > This network is a bit sparse - it is not yet SD2 complient - and
          the centrality algorithms of Ucinet have not returned satisfactory
          results.

          >D: Yet it seems somewhat clear to the human eye that nodes 7, 11 and
          24 would likely have the highest ranking (they receive the most input
          connections). Without the help of an algorithm, one would likely opt
          for these nodes as the leaders. Would that be a mistake?

          -M: Counting votes is an algorithm. In small groups, there will often
          be little difference between approval-indegree and PageRank, but even
          with this group PageRank DOES make a difference.

          The winner sequence with approval-indegree is: 11, 24, 7.

          The winner sequence with PageRank is: 24, 11, 7.

          This is because 24 is recieving 11's SOLE outgoing link, hence it is
          receiving all of 11's strength and the strength of the other incoming
          links. 11 has reduced his/her comparitive rank by voting his/her
          conscience - this is the main reason that this network is SD2 non-
          complient. With SD2, all of the top nodes MUST have 2 or more active
          outgoing links to keep this rank suicide from occuring.

          >D: Can we look forward for more results based on analysis this data?
          Specifically, PageRank? Best regards! denis

          -M: Yes, if I can get Visone to take a Ucinet or Excel file.

          -Shanti
          -Mark Rosst
          -Seattle WA USA
        Your message has been successfully submitted and would be delivered to recipients shortly.